JP3587433B2 - Pixel defect detection device for solid-state imaging device - Google Patents

Description

【００３５】
被検査画素の光電係数ａ及びオフセット出力レベルｂを最小２乗法によって求める場合、光電係数ａ及びオフセット出力レベルｂは、次式（６）によって導出される誤差の２乗の総和σを最小にすることを条件に、次式（７）に被検査画素の入射光量ｘ_ｉ及び実際の出力レベルｙ_ｉを代入すれば導出することができる（最小２乗法については、例えば日本機械学会編、朝倉書店出版、「センサと信号処理システムＩＩ」、ｐｐ１０−１１、１９８５を参照）。
【００３６】
【数２】……（６）

【００３７】
【数３】……（７）

【００３８】
本発明においては、標準光量発生装置を用いないため、被検査画素の入射光量ｘ_ｉが標準光量でなく、特定されない。このため、上記式（７）から光電係数ａ及びオフセット出力レベルｂを導出するに先立って、この入射光量ｘ_ｉを推定する。
【００３９】
被検査画素の入射光量ｘ_iを推定するために、まず、該被検査画素の近傍領域において、例えば該被検査画素の出力、及び該被検査画素の上下左右の各画素の出力を選択する。このとき、選択した各画素のうちに含まれる同じ種類の欠陥を有する画素の数が選択した各画素の半数以下であれば、選択した各画素の出力レベルの中央値ｙ _０（ｉ） を次式（８）のメディアンフィルタによって抽出して、正常な画素の出力レベルを求めることができる。
【００４０】
ｙ _０（ｉ） ＝ｍｅｄｉａｎ｛ｙ_１，ｙ_２，ｙ_３，ｙ_４，ｙ_５｝
＝ｙ１；ｗｈｅｎ ｙ_２＜ｙ_３＜ｙ_１＜ｙ_４＜ｙ_５ ……（８）
【００４１】
ただし、ｙ _ｋ （ｋ＝１，……，５）は近傍領域の５個の画素の出力レベルである。
【００４２】
例えば、ｙ₁が正常な画素の出力レベル、ｙ₂，ｙ₃が黒傷を有する画素の出力レベル（正常な画素の出力レベルよりも小さい）、ｙ₄，ｙ₅が白傷を有する画素の出力レベル（正常な画素の出力レベルよりも大きい）とすると、メディアンフィルタによって抽出された中央値ｙ _０（ｉ） は、正常な画素の出力レベルｙ₁となる。あるいは、近傍領域から正常な各画素のみが選択されたときには、正常な各画素の出力レベルのうちの中央値が選択される。
【００４３】
正常な画素の出力レベルｙ _０（ｉ） を求めると、この出力レベルｙ _０（ｉ） を次式（２）に代入して、被検査画素の入射光量ｘ _ｉ を求めることができる。
【００４４】
ｘ _ｉ ＝（ｙ _０（ｉ） −ｂ_０）／ａ_０ …（２）
ただし、ａ_０は正常な各画素の平均的な基準光電係数、ｂ_０は正常な各画素の平均的な基準オフセット出力レベルである。
【００４５】
暗から明までの相互に異なるＮ個の入射光量ｘ_０，ｘ_１，…，ｘ_{N −１}の光を入射する度に、近傍領域から正常な画素の出力レベルｙ _０（ｉ） を求め、この出力レベルｙ _０（ｉ） を上記式（２）に代入し、これによって該各入射光量ｘ_０，ｘ_１，…，ｘ_Ｎ−１を逆算して求める。
【００４６】
こうして被検査画素の各入射光量ｘ_０，ｘ_１，…，ｘ_Ｎ−１が求められると、これらの入射光量と、該各入射光量に対する被検査画素の実際の出力レベルｙ_０，ｙ_１，…，ｙ_Ｎ−１を上記式（７）に代入して、該被検査画素の光電係数ａ及びオフセット出力レベルｂを導出することができる。
【００４７】
更に、カラー表示を前提とすると、各表示色毎に、基準光電係数ａ_０が異なる。このため、例えばＲ，Ｇ，Ｂの３原色によってカラー表示を行うならば、Ｒ，Ｇ，Ｂの各原色毎に、基準光電係数ａ_０を設定する必要がある。この場合、各原色毎に、被検査画素の近傍領域から該被検査画素と同一の表示色を表す各画素を選択し、選択した各画素について、上記式（８）に基づく出力レベルの中央値の抽出、上記式（２）に基づく入射光量ｘの導出、及び上記式（７）に基づく光電係数ａの導出を逐次行う。
【００４８】
ただし、上記式（２）において、Ｒ，Ｇ，Ｂの各原色の予め設定された基準光電係数をａ_０Ｒ，ａ_０Ｇ，ａ_０Ｂとする。また、Ｒ，Ｇ，Ｂの各原色毎に求められたそれぞれの光電係数をａ_Ｒ，ａ_Ｇ，ａ_Ｂとする。
【００４９】
オフセット出力レベルｂ、及び基準オフセット出力レベルｂ_０については、各原色別に求める必要がなく、各原色に共通の値を設定すれば良い。
【００５０】
ところで、本発明の様に標準光量発生装置等の格別な機器を用いない場合は、固体撮像素子の全体に均一な光量を入射させることは非常に困難である。このため、本発明においては、被検査画素の近傍領域を可能な限り小さくして、均一な光量が入射する近傍領域を設定せねばならない。
【００５１】
例えば、カラー表示のために図２に示す様な原色ベイヤー配列のカラーフィルターを利用している場合、原色Ｒについては、被検査画素Ｒ１の近傍領域ｍａｓｋ１を設定し、この近傍領域ｍａｓｋ１から９個の各画素Ｒ１，Ｒｉを選択して、入射光量を求める。同様に、原色Ｂについては、被検査画素Ｂ１の近傍領域ｍａｓｋ２を設定し、この近傍領域ｍａｓｋ２から９個の各画素Ｂ１，Ｂｉを選択して、入射光量を求める。更に、原色Ｇについては、被検査画素Ｇ１の近傍領域ｍａｓｋ３もしくはｍａｓｋ４を設定し、これらの近傍領域ｍａｓｋ３もしくはｍａｓｋ４から９個の各画素Ｇ１，Ｇｉもしくは５個の各画素Ｇ１，Ｇｉを選択して、入射光量を求める。
【００５２】
原色Ｇは人が最も鋭敏に感知するので、この原色Ｇについては、最も小さな近傍領域mask4を設定し、これによって被検査画素の入射光量を厳密に求めるのが好ましい。
【００５３】
この様にして各被検査画素毎に、入射光量及び出力レベルを求め、被検査画素の表示色に応じて各光電係数ａ_Ｒ，ａ_Ｇ，ａ_Ｂのいずれかを導出するとともに、Ｒ，Ｇ，Ｂの各原色に共通の基準オフセット出力レベルｂ_０を導出した後、次式（９）、（１０）及び（１１）に基づいて、各被検査画素の欠陥の有無と種類を判定する。
【００５４】
被検査画素の原色Ｒを表示する場合は、

被検査画素の原色Ｂを表示する場合は、

被検査画素の原色Ｇを表示する場合は、

ただし、Δａ_Ｒ，Δａ_Ｂ，Δａ_Ｇ、及びΔｂ_Ｒ，Δｂ_Ｂ，Δｂ_Ｇはしきい値であり、判別精度の向上を考慮して、これらのしきい値をＲ，Ｇ，Ｂの各原色別に定めている。
【００５５】
なお、上記式（７）に基づいて光電係数ａ及びオフセット出力レベルｂを求めるには、該式（７）の左辺の変数が０でなければ、つまり次式（１２）が成立しなければならない。すなわち、上記式（５）のＮ個の関数列は相互に独立していなければならない。理想条件に近い（ホワイトノイズなどが存在しない）場合には、Ｎ個の関数列が相関であるために、該式（１２）が成立しなくなる。このとき、光電係数ａ＝ａ_０、オフセット出力レベルｂ＝ｂ_０となる。
【００５６】
【数５】……（１２）

【００５７】
こうして各被検査画素毎に、被検査画素の欠陥の有無と種類を判定した後、被検査画素に欠陥があれば、この欠陥を有する画素の出力レベルを補正する。
【００５８】
白傷を有する画素は、その出力に加算されるバイアス電圧が温度に応じて変化し、また黒傷を有する画素は、その感度が劣化しているので、画素の出力に対して一定の電圧を加算したり減算するという補正方法は充分とはいえず、欠陥を有する画素の出力信号を該画素の近傍の他の各画素の出力信号によって補正するのが適切である。
【００５９】
また、水平方向及び垂直方向のいずれか一方に並ぶ他の各画素の出力信号だけでなく、両方向に並ぶ他の各画素の出力信号に基づいて、欠陥を有する画素の出力信号を補正するのが望ましい。これは、欠陥を有する画素が被写体のエッジに位置している場合、エッジから水平方向及び垂直方向に並ぶ各画素による表示が異なるので、欠陥を有する画素の一方のみに並ぶ他の各画素の出力信号を用いて補正を行うと、色ずれが生じるためである。
【００６０】
更に、表示される各原色毎に、画素の出力レベルの補正を行うことが望ましい。例えばＲ、Ｇ、Ｂの各原色を表示している場合、Ｇの原色は、Ｒ、Ｂの各原色よりも重要であるため、原色ベイヤー配列の分布を利用して、Ｇの原色をより精密に補正する。
【００６１】
例えば、図２に示す様に、原色Ｒについては、被検査画素Ｒ１の近傍領域ｍａｓｋ１を設定し、原色Ｂについては、被検査画素Ｂ１の近傍領域ｍａｓｋ２を設定する。そして、被検査画素の補正された出力レベルをｙ（ｉ，ｊ）とすると、次式（１３）に基づいて、該被検査画素の近傍領域における該被検査画素と同一の表示色を有する他の８個の各画素の出力レベルの平均を求める。
【００６２】

また、図２に示す様に、原色Ｇについては、被検査画素Ｇ１の近傍領域ｍａｓｋ４を設定する。そして、被検査画素の補正された出力レベルをｙ（ｉ，ｊ）とすると、次式（１４）に基づいて、該被検査画素の近傍領域における該被検査画素と同一の表示色を有する他の４個の各画素の出力レベルの平均を求める。
【００６３】

ただし、ｉ，ｊ等は画素の座標を示す。
【００６４】
原色Ｇは人が最も鋭敏に感知するので、この原色Ｇについては、最も小さな近傍領域ｍａｓｋ４を設定し、これによって出力レベルを厳密に求めている。
【００６５】
さて、本発明の画素欠陥検出装置の一実施形態は、例えばデジタルスチルカメラの固体撮像素子（ＣＣＤ）の画素欠陥検出のために適用される。このデジタルスチルカメラには、画素欠陥検出のための専用モードを設定しておく。この専用動作モードを設定した場合は、所定の操作のもとに、固体撮像素子の欠陥検出がほぼ自動的に行われ、欠陥を有する画素の位置や該画素の出力特性が記録され、通常の撮像モードを設定したときには、この記録内容に基づいて、欠陥を有する画素の出力信号が補正される。
【００６６】
固体撮像素子の画素欠陥検出に際しては、欠陥を有する画素の出力信号の誤差量を低下させないために、固体撮像素子の出力信号に対してγ補正を施さない。
【００６７】
本実施形態の画素欠陥検出装置においては、固体撮像素子の入射光を変化させて、固体撮像素子の出力特性を求める。固体撮像素子の入射光を変化させるために、デジタルスチルカメラの絞り、ストロボ、シャッタースピード等の機能を適宜に組み合わせて、これらの機能を作動させる。
【００６８】
また、固体撮像素子の全体にほぼ一様なレベルの入射光を入射させるために、デジタルスチルカメラのピントをずらす。例えば、一様な階調の壁やパネルをデジタルスチルカメラによって撮像し、これによって固体撮像素子の全体にほぼ一様なレベルの入射光を入射させる場合でも、デジタルスチルカメラのピントをずらせば、被写体の階調ムラや該被写体に対する照明のむらをぼかすことができ、固体撮像素子の全体により均一なレベルの入射光を入射させることができる。
【００６９】
図３は、本発明の画素欠陥検出装置の一実施形態を適用したデジタルスチルカメラ示すブロック図である。図３において、光は、レンズ部１、絞り２及びシャッター３を介して固体撮像素子（例えばＣＣＤ）４に入射し、この固体撮像素子４の撮像画面に映像が投影される。固体撮像素子４は、複数の光電変換素子（以下画素と称す）を水平及び垂直方向に配列してなり、映像が該各画素に投影される。これらの画素の出力信号は、スイッチングモジュール５を介して画像処理部６に順次送出される。ストロボ７は、シャッター３の開閉動作に同期して発光し、光を被写体に照射するものである。操作キー群８は、該デジタルスチルカメラを操作するためのものである。
【００７０】
画像処理部６は、画像メモリ１１、ＥＥＲＯＭ１２、データテーブル１３及び制御信号生成部１４、プロセッサ１５等を備えている。画像メモリ１１は、固体撮像素子４の各画素の出力レベル、つまり撮像された画像を示す画像データを記録するものであり、少なくとも３つの画像を記録することができる。ＥＥＲＯＭ１２は、欠陥を有する画素の座標位置、Ｒ，Ｇ，Ｂの各原色毎に予め設定された各基準光電係数ａ_０Ｒ，ａ_０Ｇ，ａ_０Ｂ、各しきい値Δａ_Ｒ，Δａ_Ｂ，Δａ_Ｇ、各しきい値Δｂ_Ｒ，Δｂ_Ｂ，Δｂ_Ｇ、及び基準オフセット出力レベルｂ_０、固体撮像素子４の撮像画面のサイズ（撮像画面における水平及び垂直方向の各画素数Ｉ０×Ｊ０）等を記憶するものである。データテーブル１３は、画像データに施されるγ補正を行うときに用いられるデータ、画像データに施されるＪＰＥＧ圧縮を行うときに用いられるデータ等を記憶している。制御信号生成部１４は、プロセッサ１５からの指令に応答して、レンズ部１、絞り２、シャッター３、固体撮像素子４、スイッチングモジュール５、及びストロボ７等を制御するための制御信号を生成して出力する。プロセッサ１５は、該画像処理部６を統括的に制御し、画像データの処理、各種の演算処理等を行う。
【００７１】
この様な画像処理部６は、１チップのＬＳＩ上に作製することができる。
【００７２】
本実施形態のデジタルスチルカメラにおいては、操作キー群８を操作することによって、通常の撮影動作モードを選択することができる。
【００７３】
通常の撮影動作モードにおいては、操作キー群８が適宜に操作されると、これに応答してプロセッサ１５は、レンズ部１を駆動制御して、固体撮像素子４の撮像画面に投影される映像のピントを合わせ（オートフォーカス）、絞り２による絞り量を調節し、シャッター３を開閉し、ストロボ７をシャッター３の開閉動作に同期させて発光させる。この結果、映像が固体撮像素子４によって撮像される。プロセッサ１５は、スイッチングモジュール５を通じて固体撮像素子４から画像データを入力して、該画像データを画像メモリ１１に一旦記憶し、この画像データに対する画像処理（γ補正や画像圧縮）を施し、この画像データを図示しない記録媒体の録画機構に送出する。録画機構は、画像データを記録媒体に記録する。
【００７４】
また、本実施形態のデジタルスチルカメラにおいては、操作キー群８を操作することによって、画素欠陥検出モードを選択することができる。画素欠陥検出モードを選択した場合、プロセッサ１５は、図４のフローチャートに示す処理を実行し、これによって欠陥を有する画素を自動的に検出し、この画素の座標位置をＥＥＲＯＭ１２に記憶する。このため、格別の装置や知識は不要であり、ユーザであっても、欠陥画素の検出を行うことができる。
【００７５】
まず、プロセッサ１５は、画像メモリ１１内に３枚分の画像データの格納領域を用意し、またγ補正、画像圧縮、レンズ部１のオートフォーカスを停止し、更に該レンズ部１を駆動制御して、該レンズ部１のピントを例えば∞に合わせる（ステップ１０１，１０２）。
【００７６】
この後、プロセッサ１５は、シャッター３による開放時間を０に設定してから、固体撮像素子４の各画素の出力信号をスイッチングモジュール５を介して入力し、これらの画素の出力レベルを画像メモリ１１の１枚目の画像データの格納領域に記憶する（ステップ１０３）。
【００７７】
シャッター３による開放時間を０に設定した場合は、固体撮像素子４の各画素に対する入射光量が０であるから、該各画素の出力レベルは最も低くなる。
【００７８】
引き続いて、プロセッサ１５は、絞り２を開放に設定し、シャッター３を開閉し、ストロボ７をシャッター３の開閉動作に同期させて発光させる。そして、プロセッサ１５は、固体撮像素子４の各画素の出力信号をスイッチングモジュール５を介して入力し、これらの画素の出力レベルを画像メモリ１１の２枚目の画像データの格納領域に記憶する（ステップ１０４）。
【００７９】
このとき、シャッター３の開放時間は、固体撮像素子４がオーバーフローとなる直前の入射光量が該固体撮像素子４の各画素に入射する様に設定され、これによって該各画素の出力レベルが最も高くなる。また、先に述べた様に、一様な階調の壁やパネルを撮像すれば、固体撮像素子４により撮像された映像が略一様の階調になるので、欠陥を有する画素を検出するには好ましい。その上、レンズ部１のピントを∞に合わせているので、固体撮像素子４により撮像された映像がぼけて更に一様な階調となる。この様な一様の階調の映像は、出力レベルが特に低かったり（暗い）、出力レベルが特に高い（明るい）画素、つまり欠陥を有する画素を検出するのに好適である。
【００８０】
更に、プロセッサ１５は、絞り２を充分に絞り込み、シャッター３を開閉し、ストロボ７をシャッター３の開閉動作に同期させて発光させる。そして、プロセッサ１５は、固体撮像素子４の各画素の出力信号をスイッチングモジュール５を介して入力し、これらの画素の出力レベルを画像メモリ１１の３枚目の画像データの格納領域に記憶する（ステップ１０５）。
【００８１】
絞り２を充分に絞り込み、ストロボ７を発光させて、シャッター３を開閉しているので、固体撮像素子４の各画素に対する入射光量がステップ１０３と１０４のときの各入射光量の中間になり、該各画素の出力レベルも中間となる。このときも、デジタルスチルカメラの視野がステップ１０４のときと同一であるのが好ましく、先に述べた様に、一様な階調の壁やパネルを撮像するのが良い。勿論、レンズ部１のピントを∞に合わせているので、固体撮像素子４により撮像された映像がぼけて更に一様な階調となる。
【００８２】
この様に各ステップ１０３，１０４，１０５においては、入射光量が最小のときの固体撮像素子４の各画素の出力レベル、入射光量が中間のときの固体撮像素子４の各画素の出力レベル、及び入射光量が最大のときの固体撮像素子４の各画素の出力レベルを画像メモリ１１に記憶する。
【００８３】
これによって、固体撮像素子４の各画素毎に、上記式（５）の入出力関係（Ｎ−１＝２）が実現される。
【００８４】
次に、プロセッサ１５は、固体撮像素子４の各画素のアドレスデータ、つまり座標位置（ｉ，ｊ）を順次指定し、座標位置（ｉ，ｊ）を指定する度に、指定した座標位置（ｉ，ｊ）に基づいて、該座標位置（ｉ，ｊ）にある画素がＲ，Ｇ，Ｂの各原色のいずれを表示するのかを判定し、更に該画素の関数を導出して、該画素の欠陥の有無と種類を判定する。
【００８５】
本実施形態では、図２に示す原色ベイヤー配列のカラーフィルターを利用しているので、画素の座標位置（ｉ，ｊ）に基づいて、該座標位置（ｉ，ｊ）にある画素がＲ，Ｇ，Ｂの各原色のいずれを表示するのかを判定することができる。
【００８６】
最初、プロセッサ１５は、座標位置（ｉ，ｊ）を（０，０）に初期設定する。また、Ｒ，Ｇ，Ｂの各原色毎に予め設定された各基準光電係数ａ_０Ｒ，ａ_０Ｇ，ａ_０Ｂ、各しきい値Δａ_Ｒ，Δａ_Ｂ，Δａ_Ｇ、各しきい値Δｂ_Ｒ，Δｂ_Ｂ，Δｂ_Ｇ、及び基準オフセット出力レベルｂ_０、固体撮像素子４の撮像画面のサイズ（撮像画面における水平及び垂直方向の各画素数（Ｉ０−１）×（Ｊ０−１））等をＥＥＲＯＭ１２から読み出す（ステップ１０６）。
【００８７】
そして、プロセッサ１５は、座標位置（ｉ，ｊ）のｉ，ｊ別に、偶数であるか否かを判定する（ステップ１０７，１０８）。
【００８８】
ここでは、初期設定された座標位置（０，０）の０が偶数とみなされ（ステップ１０７，１０８共にＹｅｓ）、図２に示す原色ベイヤー配列においては ｉ，ｊが共に偶数である座標位置の画素がＲの原色を表示するので、Ｒの原色を表示する画素の欠陥の有無及び種類の判定が行われる（ステップ１０９）。
【００８９】
上記ステップ１０９において、プロセッサ１５は、座標位置（０，０）の画素を被検査画素とみなし、各ステップ１０３〜１０５において画像メモリ１１に記憶した３枚分の画像データから必要な各画素の出力レベルを抽出しつつ、上記式（８）及び（２）に基づいて、該被検査画素の入射光量を求め、上記式（７）に基づいて、該被検査画素の光電係数ａ及びオフセット出力レベルｂを求め、上記式（９）に基づいて、該被検査画素の欠陥の有無と種類を判定し、該被検査画素に欠陥があれば、該被検査画素の欠陥の種類（白傷又は黒傷）と該座標位置を画像メモリ１１に記憶する。
【００９０】
この後、プロセッサ１５は、ｉに１を加算して更新した座標位置（ｉ，ｊ）が画像サイズ（Ｉ０−１）から外れないことを確認した後（ステップ１１０，Ｎｏ）、ｉに１を加算して座標位置（ｉ，ｊ）を更新し（ステップ１１１）、ステップ１０７に戻る。
【００９１】
更新した座標位置（ｉ，ｊ）のｉは、１という奇数となる。従って、ｉが奇数と判定され（ステップ１０７，Ｎｏ）、ｊが偶数であると判定され（ステップ１１２，Ｙｅｓ）、図２に示す原色ベイヤー配列においては該座標位置（ｉ，ｊ）の画素がＧの原色を表示するので、Ｇの原色を表示する画素の欠陥の有無及び種類の判定が行われる（ステップ１１３）。
【００９２】
ステップ１１３において、プロセッサ１５は、座標位置（１，０）の画素を被検査画素とみなし、各ステップ１０３〜１０５において画像メモリ１１に記憶した３枚分の画像データから必要な各画素の出力レベルを抽出しつつ、上記式（８）及び（２）に基づいて、該被検査画素の入射光量を求め、上記式（７）に基づいて、該被検査画素の光電係数ａ及びオフセット出力レベルｂを求め、上記式（１０）に基づいて、該被検査画素の欠陥の有無と種類を判定し、該被検査画素に欠陥があれば、該被検査画素の欠陥の種類（白傷又は黒傷）と該座標位置を画像メモリ１１に記憶する。
【００９３】
以降同様に、図２に示す原色ベイヤー配列においては座標位置（ｉ，０）の各画素の表示色がＲとＧに交互に代わるので、ステップ１０９と１１３が交互に行われて、座標位置（ｉ，０）の各画素の欠陥の有無と種類が判定され、その度に欠陥を有する被検査画素の欠陥の種類と座標位置が画像メモリ１１に記憶される。
【００９４】
そして、ｉに１を加算して更新したｉがＩ０に達すると（ステップ１１０，Ｙｅｓ）、プロセッサ１５は、ｊに１を加算して更新した座標位置（ｉ，ｊ）が画像サイズ（Ｊ０−１）から外れないことを確認した後（ステップ１１４，Ｎｏ）、ｊに１を加算してｊを１に更新する（ステップ１１５）。このｊが奇数であるため（各ステップ１０８，１１２共にＮｏ）、ｉが偶数であるか否かに応じて（ステップ１０７）、各ステップ１１３，１１６のいずかが行われる。
【００９５】
ステップ１１３においては、先に述べた様にＧの原色を表示する画素の欠陥の有無及び種類の判定が行われる。ステップ１１４においては、Ｂの原色を表示する画素の欠陥の有無及び種類の判定が行われる。この判定を行うために、プロセッサ１５は、各ステップ１０３〜１０５において画像メモリ１１に記憶した３枚分の画像データから必要な各画素の出力レベルを抽出しつつ、上記式（８）及び（２）に基づいて、被検査画素の入射光量を求め、上記式（７）に基づいて、該被検査画素の光電係数ａ及びオフセット出力レベルｂを求め、上記式（１１）に基づいて、該被検査画素の欠陥の有無と種類を判定し、該被検査画素に欠陥があれば、該被検査画素の欠陥の種類（白傷又は黒傷）と該座標位置を画像メモリ１１に記憶する。
【００９６】
図２に示す原色ベイヤー配列から明らかな様に、座標位置（ｉ，１）の各画素の表示色がＢとＧに交互に代わるので、ステップ１０９と１１３を交互に行うことによって、座標位置（ｉ，１）の各画素の欠陥の有無と種類を判定することができる。プロセッサ１５は、欠陥を有する被検査画素の欠陥の種類と座標位置を画像メモリ１１に記憶する。
【００９７】
以降同様に、ｊに１を加算する度に、ｉを０〜（Ｉ０−１）の範囲で順次変化させて、座標位置（０〜（Ｉ０−１），ｊ）の各画素の欠陥の有無と種類を逐次判定し、１を加算して更新したｉが（Ｉ０−１）に達し（ステップ１１０，Ｙｅｓ）、かつ１を加算して更新したｊが （Ｊ０−１）に達すると（ステップ１１４，Ｙｅｓ）、全ての画素について欠陥の有無と種類の判定を終了したことになるので、プロセッサ１５は、欠陥を有する全ての各画素の欠陥の種類と座標位置を画像メモリ１１から読み出してＥＥＲＯＭ１２に記憶する。
【００９８】
こうして固体撮像素子４の全ての各欠陥画素の欠陥の種類と座標位置をＥＥＲＯＭ１２に記憶した後には、先に述べた撮影動作モードにおいて、各欠陥画素の出力レベルを補正するために次の様な処理がなされる。
【００９９】
操作キー群８を操作して、通常の撮像モードを設定すると、映像が固体撮像素子４によって撮像され、プロセッサ１５は、スイッチングモジュール５を通じて固体撮像素子４から画像データを入力する。
【０１００】
スイッチングモジュール５は、図５に示す様に構成されており、３つの各端子ＳＷ１，ＳＷ２，ＳＷ３を備えている。切片５ａが端子ＳＷ１に切換接続されているときには、固体撮像素子４の出力がそのままプロセッサ１５に送出され、各画素の出力レベルが画像メモリ１１に記憶される。
【０１０１】
また、プロセッサ１５は、ＥＥＲＯＭ１２から欠陥を有する画素の座標位置（ｉ，ｊ）を読み出し、該座標位置（ｉ，ｊ）に基づいて欠陥を有する該画素の表示色を識別し、該画素の出力信号が出力されるタイミングで、スイッチングモジュール５を端子ＳＷ２又はＳＷ３に切り換える。つまり、図４の各ステップ１０７，１０８，１１２と同様に座標位置（ｉ，ｊ）に基づいて欠陥を有する画素の表示色を判定し、該画素の表示色がＲ及びＢであれば、出力信号が該画素から出力されるタイミングで、スイッチングモジュール５を端子ＳＷ２に切り換え、また該画素の表示色がＧであれば、出力信号が該画素から出力されるタイミングで、スイッチングモジュール５を端子ＳＷ３に切り換える。
【０１０２】
プロセッサ１５は、Ｒ及びＢの原色を有する欠陥画素の出力信号をスイッチングモジュール５の端子ＳＷ２から入力したときには、上記式（１３）の演算を行うのに必要なデータが固体撮像素子４から画像メモリ１１に送出され記録された時点で、該式（１３）の演算を行って、該欠陥画素の出力レベルを求め、この求められた出力レベルを画像メモリ１１の該欠陥画素のアドレスに記憶する。
【０１０３】
また、プロセッサ１５は、Ｇの原色を有する欠陥画素の出力信号をスイッチングモジュール５の端子ＳＷ３から入力したときには、上記式（１４）の演算を行うのに必要なデータが固体撮像素子４から画像メモリ１１に送出され記録された時点で、該式（１４）の演算を行って、該欠陥画素の出力レベルを求め、この求められた出力レベルを画像メモリ１１の該欠陥画素のアドレスに記憶する。
【０１０４】
この様にスイッチングモジュール５の切片５ａを各端子ＳＷ１，ＳＷ２，ＳＷ３に切り換えることによって、固体撮像素子４からの画像データを画像メモリ１１に記憶しつつ、ほぼリアルタイムで、欠陥画素の出力レベルを補正することができる。
【０１０５】
この後、画像メモリ１１内の画像データに対する画像処理（γ補正や画像圧縮）を施し、この画像データを記録媒体の録画機構に送出して、画像データを記録媒体に記録する。
【０１０６】
なお、図４の各ステップ１０７，１０８，１１２と同様に座標位置（ｉ，ｊ）に基づいて欠陥を有する画素の表示色を判定する代わりに、ｉ及びｊの最下位ビットが０のときにはｉ及びｊが偶数であり、ｉ及びｊの最下位ビットが１のときにはｉ及びｊが奇数であることに着目し、ｉとｊの排他的論理和が偽であれば、画素の表示色がＲ又はＢであると判定し、ｉとｊの排他的論理和が真であれば、画素の表示色がＧであると判定しても良い。
【０１０７】
なお、先に述べた画素欠陥検出モードを選択したときには、切片５ａが端子ＳＷ１に常に接続されている。
【０１０８】
以上の説明から明らかな様に、本実施形態のデジタルスチルカメラにおいては、標準光量発生装置や格別の支援システム等を必要とせず、複数回の撮像によって３枚の画像データを得るだけで、欠陥画素の座標位置や種類、あるいは欠陥画素の出力信号の補正が自動的に行われるので、ユーザであっても欠陥画素の検出及び補正を容易に実施することが可能である。
【０１０９】
本発明は、上記実施形態に限定されるものでなく、多様に変形することが可能である。例えば、各画素毎に、３つの入射光量に対する３つの出力レベルに基づいて、光電係数ａ及びはオフセット出力レベルｂを求めているが、２つの入射光量に対する２つの出力レベルに基づいて、あるいは４つの入射光量に対する４つの出力レベルに基づいて、光電係数ａ及びはオフセット出力レベルｂを求めても良い。
【０１１０】
また、２つの入射光量に対する２つの出力レベルに基づいて、光電係数ａ及びはオフセット出力レベルｂを求める場合は、上記各（５），（６），（７）を必要とせず、上記式（８），（２）に基づいて２つの入射光量を求め、これらの入射光量に対する画素の２つの出力レベルを検出すれば、光電係数ａ及びオフセット出力レベルｂを導出することができる。
【０１１１】
更に、本実施形態においては、画素の出力を１次関数によって近似的に表しているが、これは演算回路の規模、演算量及び演算時間を実用的な範囲におさめるためである。光電変換素子の出力特性は、厳密には非線形である。また、光電変換素子の出力特性を１つの線形関数で近似することが困難なときには、演算回路の規模、演算量及び演算時間の増大を可能な限り抑えることを条件に、複数の線形関数の組み合わせや別の関数を適用することが考えられる。どの様な関数を採用するにしても、画素の実際の出力特性を関数で表して、この関数の係数の大小によって該画素の欠陥の有無と種類を判定する。
【０１１２】
また、上記式（８）のメディアンフィルタだけでなく、他の方法によって正常な画素の出力レベルを求めても良い。例えば、被検査画素の近傍領域における各画素の出力レベルの最大値と最小値を除いてから、メディアンフィルタを適用したり、被検査画素の近傍領域における各画素の出力レベルに対して周知の統計処理を施すことによって、正常な画素の出力レベルを求めても良い。更に、被検査画素の近傍領域として様々な領域を指定しても構わない。
【０１１３】
また、本実施形態では、ＣＣＤを例示しているが、本発明は、これに限定されるものではない。例えば、ＣＩＤ、ＣＰＤ等の固体撮像素子にも、本発明を適用することができる。あるいは、本発明は、デジタルスチルカメラだけでなく、ビデオカメラ、フィルムスキャナー等の固体撮像素子に適用することができる。
【０１１４】
更に、カラーフィルターとして、原色ベイヤー配列のカラーフィルターを例示しているが、各原色や各補色等を所定の法則に従って配列した他の種類のカラーフィルターを適用することが可能である。
【０１１５】
【発明の効果】
以上説明した様に、本発明によれば、被検査光電変換素子の入射光量を変化させたときの該入射光量に対する該被検査光電変換素子の出力特性を求め、該被検査光電変換素子の出力特性に基づいて該被検査光電変換素子の欠陥を判定している。
【０１１６】
あるいは、本発明によれば、相互に異なる複数の入射光量に対する被検査光電変換素子の各出力を画像メモリに記憶しておき、該各入射光量、該各出力及び次式（１）に基づいて、該被検査光電変換素子の光電係数ａ及び入射光量が無い状態での該被検査光電変換素子のオフセット出力レベルｂを求め、該光電係数ａ及び該オフセット出力レベルｂを予め設定された基準光電係数ａ_０及び基準オフセット出力レベルｂ_０と比較することにより、該被検査光電変換素子の欠陥を判定している。
【０１１７】
ｙ（ｘ）＝ａｘ＋ｂ …（１）
ただし、ｙ（ｘ）は前記被検査光電変換素子の出力、ｘは入射光量である。
【０１１８】
この様に被検査光電変換素子の出力特性を求め、この出力特性に基づいて、該被検査光電変換素子の欠陥を判定する場合は、従来の様に標準光量発生装置や格別の支援システム等をを必要としないので、ユーザであっても容易に光電変換素子の欠陥を検出することができる。また、光電変換素子の欠陥の有無だけでなく、その種類をも判定することが可能となる。
【０１１９】
また、固体撮像素子に対する焦点をずらした状態で、光電変換素子の出力を求めているので、該光電変換素子の近傍領域においては略一様な光が入射することになり、この近傍領域の複数の光電変換素子の出力信号から正常な光電変換素子の出力信号を特定して、実際の入射光量を推定することが可能になる。
【０１２０】
更に、各入射光量は、固体撮像素子に光が入射していないときの入射光量、及び固体撮像素子がオーバーフローとなる直前の入射光量等に設定している。これらの入射光量は、ビデオカメラやデジタルスチルカメラのシャッター速度、絞り、ストロボ等を適宜に制御することによって容易に実施することができる。
【０１２１】
また、カラー表示の場合は、各表示色別に、光電変換素子の検査を行っているので、光電変換素子の欠陥の有無と種類を正確に判定することができる。
【０１２２】
更に、カラー表示の場合は、各画素のアドレスデータ（座標位置）に基づいて、各画素の表示色を判定しているので、演算を速やかに行うことができる。
【０１２３】
【図面の簡単な説明】
【図１】光電変換素子の入出力特性を説明するために用いたグラフである。
【図２】原色ベイヤー配列のカラーフィルターにおける各原色の配列を示す平面図である。
【図３】本発明の画素欠陥検出装置の一実施形態を適用したデジタルスチルカメラ示すブロック図である。
【図４】図３の装置における処理を示すフローチャートである。
【図５】図３の装置におけるスイッチングモジュールを示すブロック図である。
【図６】光電変換素子の入出力特性を説明するために用いたグラフである。
【符号の説明】
１ レンズ部
２ 絞り
３ シャッター
４ 固体撮像素子
５ スイッチングモジュール
６ 画像処理部
７ ストロボ
８ 操作キー群
１１ 画像メモリ
１２ ＥＥＲＯＭ
１３ データテーブル
１４ 制御信号生成部
１５ プロセッサ[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a pixel defect detection device of a solid-state imaging device for automatically detecting a defect generated in a solid-state imaging device such as a CCD.
[0002]
[Prior art]
Generally, it is known that a solid-state imaging device such as a CCD causes image quality deterioration due to local crystal defects or the like (referred to as flaws) generated in a production stage. Alternatively, after shipment of the solid-state imaging device, a new flaw may be generated in the solid-state imaging device due to irradiation of cosmic rays. Such flaws are classified into two types: white flaws and black flaws.
[0003]
The graph of FIG. 6 shows the output characteristics of the photoelectric conversion element (hereinafter, referred to as a pixel) of the solid-state imaging device with respect to the amount of incident light. The solid line A shows the output characteristics of a normal pixel, and the dotted line B shows the pixel having a white defect. And the dashed line C indicates the output characteristic of a pixel having a black defect. As is clear from the output characteristics of the dotted line B, the white flaw is a defect that the bias voltage is always added to the output of the pixel. Further, as is clear from the output characteristics of the dashed line C, the black defect is a defect that the sensitivity of the pixel is deteriorated.
[0004]
Conventionally, various devices or methods for detecting these flaws and correcting the output of the pixel have been proposed.
[0005]
For example, Japanese Patent Application Laid-Open No. 8-195909 discloses a ROM for storing the position and level data of a CCD flaw already existing at the time of shipment from a factory, and a position and level data of a CCD flaw newly generated after shipment from a factory. There is disclosed a technology that includes an EEROM to be stored and corrects the output of a CCD having a flaw based on data in the ROM and the EEROM.
[0006]
Japanese Patent Application Laid-Open No. 7-7675 discloses that performing gamma correction reduces the amount of error in the CCD output signal due to flaws, making it difficult to detect flaws. There is disclosed a technique of detecting a flaw after performing inverse γ correction.
[0007]
Further, Japanese Patent Application Laid-Open No. 62-8666 discloses that an output signal of a pixel having a defect is replaced by an output signal of another pixel in a horizontal direction or an output signal of another pixel in a vertical direction. Is disclosed.
[0008]
[Problems to be solved by the invention]
However, conventionally, in order to detect the damage of the solid-state imaging device regardless of before and after shipment from the factory, a standard light amount generation device or a support device that generates reference incident light from a dark level to a bright level to the solid-state imaging device is provided. A system or the like is required, and it is difficult for a non-professional operator to detect a pixel flaw and correct a pixel output, and the implementation by a user has not been expected very much.
[0009]
Therefore, an object of the present invention is to provide a pixel defect detection device of a solid-state imaging device that does not require a standard light amount generation device or a special support system, and can be easily implemented even by a user. .
[0013]
[Means for Solving the Problems]
The pixel defect detection device of the solid-state imaging device of the present invention,In a pixel defect detection device for a solid-state imaging device in which a plurality of photoelectric conversion elements are arranged,N (N is an integer of 2 or more) lights having different amounts of light are irradiated on the photoelectric conversion element to be inspected and a plurality of photoelectric conversion elements included in the vicinity thereof, and each of the light is irradiated on the photoelectric conversion element.Inspection photoelectric conversion element for incident lightAnd the outputs of a plurality of photoelectric conversion elements included inAn image memory for storing,For each of the N incident lights, the median value y of the outputs of the plurality of photoelectric conversion elements included in the photoelectric conversion element to be inspected and the vicinity thereof stored in the image memory. _{0 (i)} (I is an integer of 0 ≦ i ≦ N−1), and the median y of the obtained N outputs is obtained. _{0 (i)} Are respectively substituted into the following equation (2) as an output signal of a photoelectric conversion element serving as a reference, and a reference photoelectric coefficient a preset in the following equation (2) is used. ₀ And reference offset output level b ₀ And substitute the light quantity x of each of the N incident lights. _i , And the obtained N incident light amounts x _i And N incident light quantities x _i N actual output signals y of the photoelectric conversion element under test for each of _i ( x ) And the photoelectric coefficient a and the following equation (1):Offset output level b of the photoelectric conversion element under inspection without incident light quantityWhen,I was askedThe photoelectric coefficient a and the offset output level bSaidReference photoelectric coefficient a₀And reference offset output level b₀Calculating means for determining the presence / absence of a pixel defect of the photoelectric conversion element to be inspected by comparingWhenIs provided.
[0014]
y _i (x) =a x _i + B (1)
x _i = (Y _{0 (i)} -B ₀ ) / A ₀ ... (2).
[0015]
In one embodiment, the optical device includes an optical unit that projects an image on the solid-state imaging device,The image memory,In a state where the focus of the optical unit is shifted with respect to the solid-state imaging device,Based on an image projected on the solid-state imaging device, the plurality of photoelectric conversion elements included in the inspection-target photoelectric conversion element and an area in the vicinity thereofOutputRemember.
[0016]
In one embodiment, the aboveNThe incident light amount isThe photoelectric conversion element to be inspected and a plurality of photoelectric conversion elements included in the vicinity areaIncident light when no light is incident onWhen,SaidInspection photoelectric conversion element and a plurality of photoelectric conversion elements included in the vicinity areaIncident light amount just before overflow occursWhenincluding.
[0017]
In one embodiment, a preset reference photoelectric conversion coefficient a₀And reference offset level b₀And an output signal y of the photoelectric conversion element serving as a reference for the photoelectric conversion element to be inspected stored in the image memory.₀Is substituted into the following equation (2) to determine the incident light amount x.
[0018]
x = (y₀-B₀) / A₀        … (2)
In one embodiment, the output y of the photoelectric conversion element₀Is set to at least the median value of the outputs of the plurality of photoelectric conversion elements included in the region near the inspection target photoelectric conversion element.
[0019]
In one embodiment, as each of the photoelectric conversion elements included in the neighboring area, only one that represents the same display color as the photoelectric conversion element to be inspected among the display colors for color display is selected.
[0020]
In one embodiment, the photoelectric coefficient a and the offset output level b of the photoelectric conversion element to be inspected and the reference photoelectric coefficient a₀And the reference offset output level b₀Is substituted into the following expression (3) to determine whether or not the photoelectric conversion element to be inspected has a defect.
[0021]
| A₀−a | <Δa and | b₀−b | <Δb: no defect ... (3)
Here, Δa and Δb are preset thresholds.
[0022]
In one embodiment, the photoelectric coefficient a and the offset output level b of the photoelectric conversion element to be inspected and the reference photoelectric coefficient a₀And the reference offset output level b₀Is substituted into the following equation (4) to determine the presence / absence and type of the defect of the photoelectric conversion element to be inspected.
[0023]
| A₀−a | <Δa and | b₀−b | <Δb: no defect
| A₀−a | ≧ Δa: With black scratch
| B₀−b | ≧ Δb: white scratches
… (4)
Here, Δa and Δb are preset thresholds.
[0024]
In one embodiment, for each display color for color display, the reference photoelectric coefficient a₀And the threshold value Δa are set.
[0025]
In one embodiment, the apparatus further includes a determination unit that determines a display color of the photoelectric conversion element to be inspected based on address data of the photoelectric conversion element to be inspected, and the reference photoelectric coefficient a is determined based on the determination by the determination unit.₀And the threshold value Δa are set.
[0026]
BEST MODE FOR CARRYING OUT THE INVENTION
First, the outline of the present invention will be described.
[0027]
As in the past, the standard light amount is supplied from the standard light amount generator to the entire solid-state image sensor, and a defective pixel is detected from the output signal of the solid-state image sensor using a special support system, and the output of this pixel is corrected. If so, the implementation by the user was not very promising.
[0028]
In the present invention, it is assumed that a standard light amount generating device for providing a standard light amount is not used. In this case, since the amount of light incident on the solid-state imaging device cannot be specified, a defective pixel cannot be identified simply by comparing the output signal of the pixel with the output signals of other pixels in the peripheral area of the pixel. . For example, as shown in the graph of FIG. 1, when the output characteristic f1 (x) of a normal pixel with respect to the amount of incident light is compared with the output characteristic f2 (x) of a pixel having a black defect with respect to the amount of incident light, the output characteristic f1 ( Since the difference between x) and the output characteristic f2 (x) (eg, Δy1, Δy2) changes according to the amount of incident light, if the amount of incident light cannot be accurately specified, a pixel having a defect is identified based on the difference. Is impossible. Further, even if only one difference between the output characteristics f1 (x) and the output characteristics f2 (x) is obtained to determine the presence or absence of a defect, white or black flaws cannot be accurately identified, and the presence or absence of a defect is not obtained. Cannot be determined exactly.
[0029]
Therefore, in the present invention, the output characteristics of the pixel of the solid-state imaging device are represented by a function of the following equation (1), and the pixel having a defect is identified using this function.
[0030]
y (x) = ax + b (1)
Here, y (x) is the output level of the pixel under inspection (photoelectric conversion element under inspection), x is the incident light amount, a is the photoelectric coefficient of the pixel, and b is the offset output level of the pixel when there is no incident light amount. is there.
[0031]
Comparing the function of each normal pixel, the values of the coefficients a and b of each function are close to each other, and the average reference photoelectric coefficient a₀, Reference offset output level b₀Becomes approximately equal to In the case of a function of a pixel having a black defect, the photoelectric coefficient a of the function of the pixel is the reference photoelectric coefficient a.₀Smaller than. Further, in the case of a function of a pixel having a white defect, the offset output level b of the function of the pixel is equal to the reference offset output level b.₀Larger than.
[0032]
Since the photoelectric coefficient a and the offset output level b of the function of a pixel are always constant irrespective of the amount of incident light, if the function of the pixel is derived, the presence / absence and type of a defect of the pixel can be specified.
[0033]
N different amounts of incident light x from dark to bright₀, X₁, ..., x_N-1To give each output level y of the pixel under test.₀, Y₁, ..., y_N-1Is obtained, the input / output relationship of the inspected pixel can be expressed by the following equation (5) based on the above equation (1).
[0034]
[Equation 1] (5)

[0035]
When the photoelectric coefficient a and the offset output level b of the pixel to be inspected are obtained by the least square method, the photoelectric coefficient a and the offset output level b minimize the sum σ of the squares of the error derived by the following equation (6). Under the condition, the incident light amount x of the pixel to be inspected is expressed by the following equation (7)._iAnd the actual output level y_i(For the least square method, see, for example, edited by The Japan Society of Mechanical Engineers, published by Asakura Shoten, “Sensor and Signal Processing System II”, pp. 10-11, 1985).
[0036]
[Formula 2] (6)

[0037]
## EQU3 ## (7)

[0038]
In the present invention, since the standard light amount generator is not used, the incident light amount x_iIs not a standard light amount and is not specified. Therefore, prior to deriving the photoelectric coefficient a and the offset output level b from the above equation (7), the incident light amount x_iIs estimated.
[0039]
Incident light amount x of pixel under inspection_iFirst, for example, the output of the pixel to be inspected and the outputs of the pixels above, below, left, and right of the pixel to be inspected are selected in the area near the pixel to be inspected. At this time, if the number of pixels having the same type of defect included in each of the selected pixels is equal to or less than half of the selected pixels, the median of the output levels of the selected pixelsy _{0 (i)} Is extracted by the median filter of the following equation (8), and the output level of a normal pixel can be obtained.
[0040]
y _{0 (i)} = Median @ y₁, Y₂, Y₃, Y₄, Y₅｝
= Y1; when y₂<Y₃<Y₁<Y₄<Y₅    …… (8)
[0041]
However,y _k (k= 1,..., 5) are output levels of five pixels in the vicinity area.
[0042]
For example, y₁Is the output level of a normal pixel, y_Two, Y_ThreeIs the output level of a pixel having a black defect (smaller than the output level of a normal pixel), y_Four, Y_FiveIs the output level of a pixel having a white defect (greater than the output level of a normal pixel), and the median value extracted by the median filtery _{0 (i)} Is the output level y of the normal pixel₁It becomes. Alternatively, when only normal pixels are selected from the neighboring area, the median of the output levels of the normal pixels is selected.
[0043]
Normal pixel output levely _{0 (i)} , This output levely _{0 (i)} Into the following equation (2) to obtain the incident light amount of the pixel to be inspected.x _i Can be requested.
[0044]
x _i = (y _{0 (i)} -B₀) / A₀        … (2)
Where a₀Is the average reference photoelectric coefficient of each normal pixel, b₀Is the average reference offset output level of each normal pixel.
[0045]
N different incident light quantities x from dark to bright₀, X₁, ..., x_{N -1}Output level of a normal pixel from the neighboring area every timey _{0 (i)} , This output levely _{0 (i)} Is substituted into the above equation (2), whereby the respective incident light amounts x₀, X₁, ..., x_N-1Is calculated backward.
[0046]
Thus, each incident light amount x of the pixel to be inspected₀, X₁, ..., x_N-1Are obtained, these incident light amounts and the actual output level y of the inspected pixel with respect to each of the incident light amounts₀, Y₁, ..., y_N-1Can be substituted into the above equation (7) to derive the photoelectric coefficient a and the offset output level b of the pixel under test.
[0047]
Further, assuming a color display, a reference photoelectric coefficient a₀Are different. For this reason, for example, if color display is performed using three primary colors of R, G, and B, the reference photoelectric coefficient a is set for each of the primary colors of R, G, and B.₀Need to be set. In this case, for each primary color, each pixel representing the same display color as the pixel under test is selected from the area near the pixel under test, and the median of the output level based on the above equation (8) is selected for each of the selected pixels. , The derivation of the incident light quantity x based on the above equation (2), and the derivation of the photoelectric coefficient a based on the above equation (7) are sequentially performed.
[0048]
However, in the above equation (2), the preset reference photoelectric coefficient of each of the primary colors R, G, and B is a_0R, A_0G, A_0BAnd Further, each photoelectric coefficient obtained for each of the primary colors R, G, and B is a_R, A_G, A_BAnd
[0049]
Offset output level b and reference offset output level b₀Does not need to be determined for each primary color, and a common value may be set for each primary color.
[0050]
By the way, when a special device such as a standard light amount generator is not used as in the present invention, it is very difficult to make a uniform light amount incident on the entire solid-state imaging device. For this reason, in the present invention, the area near the pixel to be inspected is made as small as possible to set the area where uniform light quantity is incident.I have toNo.
[0051]
For example, when a color filter having a Bayer array of primary colors as shown in FIG. 2 is used for color display, for the primary color R, a neighboring area mask1 of the pixel to be inspected R1 is set, and nine pixels from the neighboring area mask1 are set. Are selected to determine the amount of incident light. Similarly, for the primary color B, a neighboring area mask2 of the pixel B1 to be inspected is set, and nine pixels B1 and Bi are selected from the neighboring area mask2 to determine the amount of incident light. Further, for the primary color G, a neighboring area mask3 or mask4 of the inspected pixel G1 is set, and nine pixels G1 and Gi or five pixels G1 and Gi are selected from these neighboring areas mask3 or mask4. And the amount of incident light.
[0052]
Since the primary color G is most sensitively perceived by a human, the smallest neighboring area mask4 is set for the primary color G, whereby the incident light amount of the pixel to be inspected is strictly determined.AskIs preferred.
[0053]
In this way, the amount of incident light and the output level are determined for each pixel to be inspected, and each photoelectric coefficient a is determined according to the display color of the pixel to be inspected._R, A_G, A_BAnd a reference offset output level b common to each of the R, G, and B primary colors.₀Is derived, the presence / absence and type of each defect pixel are determined based on the following equations (9), (10) and (11).
[0054]
When displaying the primary color R of the pixel to be inspected,

When displaying the primary color B of the pixel to be inspected,

When displaying the primary color G of the pixel to be inspected,

Where Δa_R, Δa_B, Δa_G, And Δb_R, Δb_B, Δb_GAre threshold values, and these threshold values are determined for each of the primary colors R, G, and B in consideration of the improvement of the discrimination accuracy.
[0055]
In order to calculate the photoelectric coefficient a and the offset output level b based on the above equation (7), the variable on the left side of the equation (7) is not 0, that is, the following equation (12) must be satisfied. . That is, the N function strings in the above equation (5) must be independent of each other. When the condition is close to the ideal condition (white noise or the like does not exist), the equation (12) does not hold because the N function strings are correlated. At this time, the photoelectric coefficient a = a₀, Offset output level b = b₀It becomes.
[0056]
## EQU5 ## (12)

[0057]
After the presence or absence and type of the defect of the inspected pixel are determined for each inspected pixel in this way, if the inspected pixel has a defect, the output level of the defective pixel is corrected.
[0058]
For a pixel having a white defect, the bias voltage added to its output changes according to the temperature, and for a pixel having a black defect, the sensitivity has deteriorated. The correction method of adding or subtracting is not sufficient, and it is appropriate to correct the output signal of a defective pixel by the output signals of other pixels near the pixel.
[0059]
In addition, based on not only the output signals of the other pixels arranged in one of the horizontal direction and the vertical direction but also the output signals of the other pixels arranged in both directions, the output signal of the defective pixel is corrected. desirable. This is because when the defective pixel is located at the edge of the subject, the display by the pixels arranged in the horizontal and vertical directions from the edge is different, so that the defective pixel is displayed.ofThis is because a color shift occurs when the correction is performed using the output signals of the other pixels arranged only on one side.
[0060]
Further, it is desirable to correct the output level of the pixel for each primary color displayed. For example, when displaying the primary colors of R, G, and B, the primary colors of G are more important than the primary colors of R and B. Therefore, the primary colors of G can be more precisely determined by using the distribution of the primary color Bayer array. To be corrected.
[0061]
For example, as shown in FIG. 2, for the primary color R, a neighboring area mask1 of the pixel to be inspected R1 is set, and for the primary color B, a neighboring area mask2 of the pixel to be inspected B1 is set. Then, assuming that the corrected output level of the pixel to be inspected is y (i, j), based on the following equation (13), another pixel having the same display color as that of the pixel to be inspected in a region near the pixel to be inspected. The average of the output levels of each of the eight pixels is obtained.
[0062]

Further, as shown in FIG. 2, for the primary color G, an area mask4 near the pixel G1 to be inspected is set. Then, assuming that the corrected output level of the pixel to be inspected is y (i, j), based on the following equation (14), another pixel having the same display color as that of the pixel to be inspected in a region near the pixel to be inspected. The average of the output levels of each of the four pixels is obtained.
[0063]

Here, i, j, etc. indicate the coordinates of the pixel.
[0064]
Since the primary color G is most sensitively sensed by a person, the smallest neighboring area mask4 is set for the primary color G, and the output level is strictly determined.
[0065]
An embodiment of the pixel defect detection device according to the present invention is applied to, for example, pixel defect detection of a solid-state imaging device (CCD) of a digital still camera. A dedicated mode for detecting a pixel defect is set in the digital still camera. When this dedicated operation mode is set, the defect detection of the solid-state imaging device is almost automatically performed under a predetermined operation, the position of the defective pixel and the output characteristics of the pixel are recorded, and the normal operation is performed. When the imaging mode is set, the output signal of the defective pixel is corrected based on the recorded contents.
[0066]
In detecting a pixel defect of the solid-state imaging device, γ correction is not performed on the output signal of the solid-state imaging device in order not to reduce the error amount of the output signal of the defective pixel.
[0067]
In the pixel defect detection device according to the present embodiment, the output characteristics of the solid-state imaging device are obtained by changing the incident light of the solid-state imaging device. In order to change the incident light of the solid-state imaging device, these functions are operated by appropriately combining the functions such as the aperture, strobe, and shutter speed of the digital still camera.
[0068]
Further, the focus of the digital still camera is shifted so that incident light of a substantially uniform level is incident on the entire solid-state imaging device. For example, even when a wall or panel with a uniform gradation is imaged by a digital still camera and incident light of a substantially uniform level is incident on the entire solid-state imaging device, if the digital still camera is out of focus, It is possible to blur the gradation unevenness of the subject and uneven illumination of the subject, and to make incident light of a uniform level incident on the entire solid-state imaging device.
[0069]
FIG. 3 is a block diagram showing a digital still camera to which an embodiment of the pixel defect detection device according to the present invention is applied. In FIG. 3, light enters a solid-state imaging device (for example, a CCD) 4 via a lens unit 1, a diaphragm 2, and a shutter 3, and an image is projected on an imaging screen of the solid-state imaging device 4. The solid-state imaging device 4 is configured by arranging a plurality of photoelectric conversion elements (hereinafter, referred to as pixels) in horizontal and vertical directions, and an image is projected on each of the pixels. Output signals of these pixels are sequentially sent to the image processing unit 6 via the switching module 5. The strobe 7 emits light in synchronization with the opening and closing operation of the shutter 3, and irradiates the subject with light. The operation key group 8 is for operating the digital still camera.
[0070]
The image processing unit 6 includes an image memory 11, an EEPROM 12, a data table 13, a control signal generation unit 14, a processor 15, and the like. The image memory 11 stores an output level of each pixel of the solid-state imaging device 4, that is, image data indicating a captured image, and can record at least three images. The EEPROM 12 stores a coordinate position of a defective pixel, a reference photoelectric coefficient a preset for each of R, G, and B primary colors._0R, A_0G, A_0B, Each threshold value Δa_R, Δa_B, Δa_G, Each threshold value Δb_R, Δb_B, Δb_G, And reference offset output level b₀, The size of the imaging screen of the solid-state imaging device 4 (the number of pixels I0 × J0 in the horizontal and vertical directions in the imaging screen) and the like. The data table 13 stores data used when performing gamma correction performed on image data, data used when performing JPEG compression performed on image data, and the like. The control signal generation unit 14 generates a control signal for controlling the lens unit 1, the aperture 2, the shutter 3, the solid-state imaging device 4, the switching module 5, the strobe 7, and the like in response to a command from the processor 15. Output. The processor 15 controls the image processing unit 6 in an integrated manner, and performs processing of image data, various arithmetic processing, and the like.
[0071]
Such an image processing unit 6 can be manufactured on a one-chip LSI.
[0072]
In the digital still camera of the present embodiment, a normal shooting operation mode can be selected by operating the operation key group 8.
[0073]
In the normal photographing operation mode, when the operation key group 8 is appropriately operated, the processor 15 controls the driving of the lens unit 1 in response to the operation, and the image projected on the imaging screen of the solid-state imaging device 4. Is adjusted (autofocus), the aperture amount of the aperture 2 is adjusted, the shutter 3 is opened and closed, and the flash 7 emits light in synchronization with the opening and closing operation of the shutter 3. As a result, an image is captured by the solid-state imaging device 4. The processor 15 inputs image data from the solid-state imaging device 4 through the switching module 5, temporarily stores the image data in the image memory 11, performs image processing (γ correction and image compression) on the image data, and The data is sent to a recording mechanism (not shown) of a recording medium. The recording mechanism records the image data on a recording medium.
[0074]
In the digital still camera according to the present embodiment, a pixel defect detection mode can be selected by operating the operation keys 8. When the pixel defect detection mode is selected, the processor 15 executes the processing shown in the flowchart of FIG. 4, thereby automatically detecting a defective pixel, and stores the coordinate position of the pixel in the EEPROM 12. Therefore, no special device or knowledge is required, and even a user can detect a defective pixel.
[0075]
First, the processor 15 prepares a storage area for three pieces of image data in the image memory 11, stops γ correction, image compression, and autofocus of the lens unit 1, and further controls the driving of the lens unit 1. Then, the focus of the lens unit 1 is adjusted to, for example, ∞ (steps 101 and 102).
[0076]
Thereafter, the processor 15 sets the opening time of the shutter 3 to 0, inputs the output signals of the respective pixels of the solid-state imaging device 4 via the switching module 5, and stores the output levels of these pixels in the image memory 11 Is stored in the storage area for the first image data (step 103).
[0077]
When the opening time of the shutter 3 is set to 0, the amount of light incident on each pixel of the solid-state imaging device 4 is 0, so that the output level of each pixel becomes the lowest.
[0078]
Subsequently, the processor 15 sets the aperture 2 to open, opens and closes the shutter 3, and causes the strobe 7 to emit light in synchronization with the opening and closing operation of the shutter 3. Then, the processor 15 inputs the output signals of each pixel of the solid-state imaging device 4 via the switching module 5 and stores the output levels of these pixels in the storage area of the second image data of the image memory 11 ( Step 104).
[0079]
At this time, the opening time of the shutter 3 is set such that the amount of incident light immediately before the solid-state imaging device 4 overflows enters each pixel of the solid-state imaging device 4, whereby the output level of each pixel is the highest. Become. Further, as described above, if an image of a wall or panel with a uniform gradation is taken, the image taken by the solid-state imaging device 4 has a substantially uniform gradation, so that a defective pixel is detected. Is preferred. In addition, since the focus of the lens unit 1 is set to ∞, the image captured by the solid-state imaging device 4 is blurred and has a more uniform gradation. Such an image having a uniform gradation is suitable for detecting a pixel having a particularly low output level (dark) or a particularly high output level (bright), that is, a pixel having a defect.
[0080]
Further, the processor 15 sufficiently stops down the aperture 2, opens and closes the shutter 3, and causes the strobe 7 to emit light in synchronization with the opening and closing operation of the shutter 3. Then, the processor 15 inputs the output signals of each pixel of the solid-state imaging device 4 via the switching module 5 and stores the output levels of these pixels in the storage area of the third image data in the image memory 11 ( Step 105).
[0081]
Since the stop 2 is sufficiently stopped down, the strobe light is emitted, and the shutter 3 is opened and closed, the amount of incident light to each pixel of the solid-state imaging device 4 is in the middle of each amount of incident light in steps 103 and 104. The output level of each pixel is also intermediate. Also in this case, it is preferable that the field of view of the digital still camera is the same as that in step 104, and as described above, it is preferable to capture an image of a wall or panel having a uniform gradation. Of course, since the focus of the lens unit 1 is set to ∞, the image picked up by the solid-state image pickup device 4 is blurred and the gradation becomes more uniform.
[0082]
As described above, in steps 103, 104, and 105, the output level of each pixel of the solid-state imaging device 4 when the amount of incident light is the minimum, the output level of each pixel of the solid-state imaging device 4 when the amount of incident light is intermediate, and The output level of each pixel of the solid-state imaging device 4 when the amount of incident light is maximum is stored in the image memory 11.
[0083]
Thus, the input / output relationship (N-1 = 2) of the above equation (5) is realized for each pixel of the solid-state imaging device 4.
[0084]
Next, the processor 15 sequentially specifies the address data of each pixel of the solid-state imaging device 4, that is, the coordinate position (i, j), and each time the coordinate position (i, j) is specified, the specified coordinate position (i, j). , J), it is determined which of the primary colors R, G, and B the pixel at the coordinate position (i, j) displays, and a function of the pixel is further derived to determine the function of the pixel. The presence and type of the defect are determined.
[0085]
In the present embodiment, since the color filter of the primary color Bayer array shown in FIG. 2 is used, the pixel at the coordinate position (i, j) is R, G based on the coordinate position (i, j) of the pixel. , B to be displayed.
[0086]
First, the processor 15 initializes the coordinate position (i, j) to (0, 0). Each reference photoelectric coefficient a preset for each of the primary colors R, G, and B_0R, A_0G, A_0B, Each threshold value Δa_R, Δa_B, Δa_G, Each threshold value Δb_R, Δb_B, Δb_G, And reference offset output level b₀Then, the size of the imaging screen of the solid-state imaging device 4 (the number of pixels (I0-1) × (J0-1) in the horizontal and vertical directions on the imaging screen) and the like are read from the EEPROM 12 (step 106).
[0087]
Then, the processor 15 determines whether or not the coordinate position (i, j) is an even number for each of i and j (steps 107 and 108).
[0088]
Here, 0 in the initially set coordinate position (0, 0) is regarded as an even number (Yes in both steps 107 and 108), and in the primary color Bayer array shown in FIG. Since the pixel displays the primary color of R, the presence or absence and the type of the defect of the pixel displaying the primary color of R are determined (step 109).
[0089]
In step 109, the processor 15 regards the pixel at the coordinate position (0, 0) as the pixel to be inspected, and outputs the necessary pixels from the three pieces of image data stored in the image memory 11 in steps 103 to 105. While extracting the level, the incident light amount of the pixel to be inspected is obtained based on the above equations (8) and (2), and the photoelectric coefficient a and the offset output level of the pixel to be inspected are obtained based on the above equation (7). b is determined, and the presence / absence and type of the defect of the inspected pixel are determined based on the above equation (9). If the inspected pixel has a defect, the type of the defect (white defect or black defect) of the inspected pixel is determined. The flaw) and the coordinate position are stored in the image memory 11.
[0090]
Thereafter, the processor 15 confirms that the coordinate position (i, j) updated by adding 1 to i does not deviate from the image size (I0-1) (step 110, No), and then adds 1 to i. The coordinate position (i, j) is updated by the addition (step 111), and the process returns to step 107.
[0091]
I of the updated coordinate position (i, j) is an odd number of 1. Therefore, i is determined to be odd (step 107, No), j is determined to be even (step 112, Yes), and in the primary color Bayer array shown in FIG. 2, the pixel at the coordinate position (i, j) is Since the primary color of G is displayed, the presence or absence and the type of the defect of the pixel displaying the primary color of G are determined (step 113).
[0092]
In step 113, the processor 15 regards the pixel at the coordinate position (1, 0) as a pixel to be inspected, and determines the output level of each pixel required from the three pieces of image data stored in the image memory 11 in steps 103 to 105. , The incident light amount of the pixel under inspection is obtained based on the above equations (8) and (2), and the photoelectric coefficient a and the offset output level b of the pixel under inspection are obtained based on the above equation (7). Is determined based on the above equation (10), and the presence / absence and type of the defect of the inspected pixel are determined. If the inspected pixel has a defect, the type of the defect (white or black ) And the coordinate position are stored in the image memory 11.
[0093]
Similarly, in the primary color Bayer array shown in FIG. 2, since the display color of each pixel at the coordinate position (i, 0) alternates between R and G, steps 109 and 113 are performed alternately, and the coordinate position ( The presence / absence and type of a defect in each pixel (i, 0) are determined, and the type and coordinate position of the defect of the inspected pixel having the defect are stored in the image memory 11 each time.
[0094]
When i updated by adding 1 to i reaches I0 (step 110, Yes), the processor 15 adds the 1 to j and updates the coordinate position (i, j) to the image size (J0− After confirming that it does not deviate from 1) (Step 114, No), 1 is added to j to update j to 1 (Step 115). Since j is an odd number (No in each of steps 108 and 112), one of steps 113 and 116 is performed depending on whether i is an even number (step 107).
[0095]
In step 113, as described above, the presence / absence and type of the defect of the pixel displaying the G primary color are determined. In step 114, the presence / absence and type of the defect of the pixel displaying the primary color B are determined. In order to make this determination, the processor 15 extracts the required output level of each pixel from the three pieces of image data stored in the image memory 11 in each of steps 103 to 105, and performs the above equations (8) and (2). ), The incident light amount of the pixel to be inspected is determined, the photoelectric coefficient a and the offset output level b of the pixel to be inspected are determined based on the above equation (7), and the detected light quantity is calculated based on the above equation (11). The presence / absence and type of the defect of the inspection pixel are determined, and if the inspection pixel has a defect, the type (white or black defect) of the defect and the coordinate position of the inspection pixel are stored in the image memory 11.
[0096]
As is apparent from the primary color Bayer array shown in FIG. 2, the display color of each pixel at the coordinate position (i, 1) is alternately changed to B and G. Therefore, by performing steps 109 and 113 alternately, the coordinate position ( The presence / absence and type of each pixel in i, 1) can be determined. The processor 15 stores the type and coordinate position of the defect of the inspected pixel having the defect in the image memory 11.
[0097]
Similarly, every time 1 is added to j, i is sequentially changed in the range of 0 to (I0-1) to determine whether or not each pixel at the coordinate position (0 to (I0-1), j) has a defect. And the type is successively determined, and the value i updated by adding 1 reaches (I0-1) (Step 110, Yes), and the value j updated by adding 1 reaches (J0-1) (Step 110). 114, Yes), which means that the determination of the presence / absence and the type of the defect has been completed for all the pixels, so that the processor 15 reads out the type and the coordinate position of the defect of each of the pixels having the defect from the image memory 11 and reads the EEPROM 12 To memorize.
[0098]
After the defect types and coordinate positions of all the defective pixels of the solid-state imaging device 4 are stored in the EEPROM 12 in this manner, in the above-described shooting operation mode, the following operation is performed to correct the output level of each defective pixel. Processing is performed.
[0099]
When a normal imaging mode is set by operating the operation key group 8, an image is captured by the solid-state imaging device 4, and the processor 15 inputs image data from the solid-state imaging device 4 through the switching module 5.
[0100]
The switching module 5 is configured as shown in FIG. 5, and includes three terminals SW1, SW2, and SW3. When the section 5a is switched to the terminal SW1, the output of the solid-state imaging device 4 is sent to the processor 15 as it is, and the output level of each pixel is stored in the image memory 11.
[0101]
Further, the processor 15 reads the coordinate position (i, j) of the defective pixel from the EEPROM 12, identifies the display color of the defective pixel based on the coordinate position (i, j), and outputs the pixel. At the timing when the signal is output, the switching module 5 is switched to the terminal SW2 or SW3. That is, the display color of the defective pixel is determined based on the coordinate position (i, j) as in the respective steps 107, 108, and 112 in FIG. When the signal is output from the pixel, the switching module 5 is switched to the terminal SW2. When the display color of the pixel is G, the switching module 5 is switched to the terminal SW3 at the timing when the output signal is output from the pixel. Switch to.
[0102]
When the output signal of the defective pixel having the primary colors of R and B is input from the terminal SW <b> 2 of the switching module 5, the processor 15 stores data necessary for performing the operation of the above equation (13) from the solid-state imaging device 4 into the image memory. At the time when the data is sent to and recorded in the memory 11, the calculation of the equation (13) is performed to determine the output level of the defective pixel, and the calculated output level is stored in the image memory 11 at the address of the defective pixel.
[0103]
Further, when the output signal of the defective pixel having the primary color of G is input from the terminal SW3 of the switching module 5, the processor 15 stores the data necessary for performing the operation of the above equation (14) from the solid-state imaging device 4 into the image memory. At the time when the data is sent to and recorded on the defective pixel 11, the operation of the equation (14) is performed to determine the output level of the defective pixel, and the determined output level is stored in the image memory 11 at the address of the defective pixel.
[0104]
By switching the section 5a of the switching module 5 to the terminals SW1, SW2, and SW3 in this manner, the image data from the solid-state imaging device 4 is stored in the image memory 11 and the output level of the defective pixel is corrected almost in real time. can do.
[0105]
Thereafter, image processing (γ correction and image compression) is performed on the image data in the image memory 11, and the image data is sent to a recording mechanism of a recording medium, and the image data is recorded on the recording medium.
[0106]
Note that, instead of determining the display color of a defective pixel based on the coordinate position (i, j) as in the respective steps 107, 108, and 112 in FIG. 4, when the least significant bits of i and j are 0, i Note that i and j are odd when the least significant bit of i and j is 1, and if the exclusive OR of i and j is false, the display color of the pixel is R Alternatively, it may be determined that the pixel is B, and if the exclusive OR of i and j is true, the display color of the pixel may be determined to be G.
[0107]
When the above-described pixel defect detection mode is selected, the intercept 5a is always connected to the terminal SW1.
[0108]
As is apparent from the above description, the digital still camera according to the present embodiment does not require a standard light amount generating device or a special support system, and only obtains three pieces of image data by performing multiple image capturing. Since the correction of the coordinate position and type of the pixel or the output signal of the defective pixel is automatically performed, even a user can easily detect and correct the defective pixel.
[0109]
The present invention is not limited to the above embodiment, but can be variously modified. For example, for each pixel, the photoelectric coefficient a and the offset output level b are obtained based on three output levels for three incident light amounts. However, based on two output levels for two incident light amounts, or The photoelectric coefficient a and the offset output level b may be obtained based on four output levels for one incident light amount.
[0110]
When the photoelectric coefficient a and the offset output level b are obtained based on two output levels with respect to two incident light amounts, the above equations (5), (6), and (7) are not required, and the above equation ( 8) and (2), two incident light amounts are obtained, and two output levels of the pixel with respect to these incident light amounts are detected.as well asAn offset output level b can be derived.
[0111]
Furthermore, in the present embodiment, the output of the pixel is approximately represented by a linear function, in order to reduce the scale, the amount of operation, and the operation time of the operation circuit to a practical range. The output characteristics of the photoelectric conversion element are strictly non-linear. When it is difficult to approximate the output characteristics of the photoelectric conversion element with one linear function, a combination of a plurality of linear functions is provided on the condition that the increase in the scale, the amount of operation, and the operation time of the operation circuit is suppressed as much as possible. Or applying another function. Whatever function is used, the actual output characteristics of a pixel are represented by a function, and the presence / absence and type of a defect of the pixel are determined based on the magnitude of the coefficient of this function.
[0112]
In addition to the median filter of the above equation (8), the output level of a normal pixel may be obtained by another method. For example, after removing the maximum value and the minimum value of the output level of each pixel in the area near the pixel to be inspected, a median filter is applied, or a well-known statistical method is applied to the output level of each pixel in the area near the pixel to be inspected. By performing the processing, the output level of a normal pixel may be obtained. Further, various regions may be designated as the region near the pixel to be inspected.
[0113]
Further, in the present embodiment, a CCD is exemplified, but the present invention is not limited to this. For example, the present invention can be applied to solid-state imaging devices such as CID and CPD. Alternatively, the present invention can be applied to not only a digital still camera but also a solid-state imaging device such as a video camera and a film scanner.
[0114]
Further, as the color filter, a color filter of a primary color Bayer array is illustrated, but other types of color filters in which each primary color, each complementary color, and the like are arranged according to a predetermined rule can be applied.
[0115]
【The invention's effect】
As described above, according to the present invention, the output characteristics of the photoelectric conversion element under test with respect to the incident light quantity when the incident light quantity of the photoelectric conversion element under test is changed, and the output characteristics of the photoelectric conversion element under test are obtained. The defect of the inspected photoelectric conversion element is determined based on the characteristics.
[0116]
Alternatively, according to the present invention, each output of the photoelectric conversion element to be inspected for a plurality of mutually different incident light amounts is stored in an image memory, and based on each incident light amount, each output, and the following equation (1). A photoelectric coefficient a of the inspected photoelectric conversion element and an offset output level b of the inspected photoelectric conversion element in a state where there is no incident light amount, and the photoelectric coefficient a and the offset output level b are set to a predetermined reference photoelectric level. Coefficient a₀And reference offset output level b₀By comparing with, the defect of the inspected photoelectric conversion element is determined.
[0117]
y (x) = ax + b (1)
Here, y (x) is the output of the photoelectric conversion element under inspection, and x is the amount of incident light.
[0118]
In this way, when the output characteristic of the photoelectric conversion element to be inspected is obtained and the defect of the photoelectric conversion element to be inspected is determined based on the output characteristic, a standard light amount generator or a special support system or the like is used as in the related art. Is not required, so that even a user can easily detect a defect in the photoelectric conversion element. In addition, it is possible to determine not only the presence / absence of a defect in the photoelectric conversion element but also the type thereof.
[0119]
In addition, since the output of the photoelectric conversion element is obtained in a state where the focus with respect to the solid-state imaging element is shifted, substantially uniform light enters in the vicinity of the photoelectric conversion element. The normal output signal of the photoelectric conversion element can be specified from the output signal of the photoelectric conversion element, and the actual incident light amount can be estimated.
[0120]
Further, each incident light amount is set to an incident light amount when no light is incident on the solid-state imaging device, an incident light amount immediately before the solid-state imaging device overflows, and the like. These incident light amounts can be easily implemented by appropriately controlling the shutter speed, aperture, strobe, and the like of a video camera or a digital still camera.
[0121]
In the case of color display, since the inspection of the photoelectric conversion elements is performed for each display color, the presence or absence and the type of the defects of the photoelectric conversion elements can be accurately determined.
[0122]
Furthermore, in the case of color display, since the display color of each pixel is determined based on the address data (coordinate position) of each pixel, the calculation can be performed quickly.
[0123]
[Brief description of the drawings]
FIG. 1 is a graph used to explain input / output characteristics of a photoelectric conversion element.
FIG. 2 is a plan view showing an arrangement of each primary color in a color filter having a primary color Bayer arrangement.
FIG. 3 is a block diagram illustrating a digital still camera to which an embodiment of the pixel defect detection device according to the present invention is applied.
FIG. 4 is a flowchart showing a process in the apparatus of FIG. 3;
FIG. 5 is a block diagram showing a switching module in the device of FIG.
FIG. 6 is a graph used to explain input / output characteristics of a photoelectric conversion element.
[Explanation of symbols]
1 Lens section
2 Aperture
3 Shutter
4 Solid-state imaging device
5 Switching module
6 Image processing unit
7 Strobe
8 Operation keys
11 Image memory
12 EEPROM
13 Data Table
14 Control signal generator
15 Processor

Claims (8)

In a pixel defect detection device for a solid-state imaging device in which a plurality of photoelectric conversion elements are arranged,
N (N is an integer of 2 or more) lights having different light amounts are radiated to the photoelectric conversion element to be inspected and a plurality of photoelectric conversion elements included in the vicinity thereof, and each of the incident lights is inspected. An image memory for storing the outputs of the photoelectric conversion elements and the plurality of photoelectric conversion elements included in the vicinity thereof ,
For each of the N incident lights, the median value y0 _(i) of the outputs of the photoelectric conversion elements to be inspected stored in the image memory and the plurality of photoelectric conversion elements included in the vicinity thereof (i is 0 ≦ i ≦ N−1 ), and substitute the obtained median value y _{0 (i)} of the N outputs into the following equation (2) as a reference output signal of the photoelectric conversion element. to together, by substituting the reference photoelectric coefficient a ₀ and the reference offset output level b ₀ which is set in advance in the following equation (2), determine the respective light intensity x _i of N incident light, the obtained N pieces of Using the incident light quantity x _i and the actual N output signals y _i ( x ) of the photoelectric conversion element under test for each of the N incident light quantities x _i , the photoelectric coefficient is calculated based on the following equation (1). a and the offset obtaining step photoelectric conversion element in a state the amount of incident light is not Obtains a power level b, by the sought photoelectric coefficient a and the offset output level b is compared with the reference photoelectric coefficient a ₀ and the reference offset output level b _0, absence of the pixel defect obtaining step photoelectric conversion element pixel defect detecting apparatus of a solid-state imaging device and a determining operation means.
y _i (x) = a · x _i + b (1)
_{x i = (y 0 (i} ) -b 0) / a 0 ... (2). Optical means for projecting an image on the solid-state imaging device,
The image memory includes a plurality of photoelectric conversion elements included in the photoelectric conversion element to be inspected and an area in the vicinity thereof, based on an image projected on the solid-state imaging element with the optical unit being out of focus with respect to the solid-state imaging element. The pixel defect detection device for a solid-state imaging device according to claim 1 , wherein an output of the device is stored . The N amount of incident light, the incident light intensity and, the inspection photoelectric conversion element and its neighboring region of when the light into a plurality of photoelectric conversion elements included in the inspection photoelectric conversion element and its neighboring region does not enter a plurality of pixel defect detector for a solid state image pickup device according to claim 1 or 2 photoelectric conversion element and a quantity of incident light immediately before the overflow contained. As the photoelectric conversion elements included in the neighboring region, only one of the claims 1 to 3 is selected to represent the same display color and the inspection photoelectric conversion elements of the respective display colors for color display 13. A pixel defect detection device for a solid-state imaging device according to claim 1. Substituting the photoelectric coefficient a and the offset output level b of the inspected photoelectric conversion element and the reference photoelectric coefficient a ₀ and the reference offset output level b ₀ into the following equation (3), the defect of the inspected photoelectric conversion element is obtained. The pixel defect detection device for a solid-state imaging device according to claim 1 , wherein the presence or absence of the defect is determined.
| A ₀ −a | <Δa and | b ₀ −b | <Δb: no defect (3)
Here, Δa and Δb are preset thresholds. By substituting the photoelectric coefficient a and the offset output level b of the inspected photoelectric conversion element and the reference photoelectric coefficient a ₀ and the reference offset output level b ₀ into the following equation (4), the defect of the inspected photoelectric conversion element is determined. The pixel defect detection device for a solid-state imaging device according to any one of claims 1 to 4 , wherein the presence / absence and type of the pixel defect are determined.
| A ₀ -a | <Δa and | b ₀ -b | <Δb: no defect | a ₀ -a | ≧ Δa: black defect | b ₀ -b | ≧ Δb: white defect… (4)
Here, Δa and Δb are preset thresholds. For each display color for color display, the pixel defect detector for a solid state image pickup device according to claim 5 or 6 for setting the reference photoelectric coefficient a ₀ and said threshold .DELTA.a. A determination unit configured to determine a display color of the photoelectric conversion element under inspection based on address data of the photoelectric conversion element under inspection;
The pixel defect detection device for a solid-state imaging device according to claim 5 , wherein the reference photoelectric coefficient a ₀ and the threshold value Δa are set based on the determination by the determination unit.

Images